Comparative Study of Near-Lossless Compression by JPEG XR, JPEG 2000, and H.264 on 4K Video Sequences - Advances in Image and Graphics Technologies - page 30

The goal of JPEG XR is to
the encoder and the decode
formance twice as high as J
of computational cost [3].
XR coding for timed sequen
The main difference betw
ing table.

achieve state-of-the-art image compression, while keep
er complexity lower [2]. It supports high compression p
JPEG, and also has an advantage over JPEG2000 in ter
Motion JPEG XR [4] Standard specifies the use of JP
nces of images
ween JPEG XR, JPEG and JPEG 2000 is shown in follo

ping
per-
rms
PEG

ow-

Table 1. Comp parison between JPEG, JPEG 2000 and JPEG XR

Standard JP PEG JPEG 2000 JPEG XR

HDR support

× √ √

Subjective lossless
compression ratio

≈ 1/5 ≈ 1/10 ≈ 1/10

Lossy/lossless com-
pression

Lo
(JP
po

ossy
PEG LS sup-
ort lossless)

Lossy/lossless Lossy/lossless

Scalability best Good

×

Key technology Di

Tr
(D
en

iscrete Cosine
ransform
DCT), Huffman
tropy coding

Discrete Wavelet
Transform(DWT),
adaptive entropy
coding

Photo Co
Transform(PCT
adaptive entrop
coding

ore
T),
py

Encoder complexity 1(b base) 15 times 8 times

Like JPEG 2000, an ima
one by one more easily in
presented to divide an imag

age is divided into small parts so that they can be proces
JPEG XR standard. As in Figure 1, three hierarchies
ge. They are tiles, macroblocks and blocks.

ssed
are

Fig. 1. An exam mple of hierarchical image partitioning in JPEG XR

Tile size has to be decid
largely related to the hardw
efficiency is explored in Se

ded at the beginning of JPEG XR compression since i
ware architecture [5]. The influence of tile size on cod
ction 2.2

it is
ding